Neural networks are complex arrangements of neurons with a variety of dynamic intrinsic properties, and exist in dynamic physiological environments. Achieving stable function requires that these networks regulate their properties to preserve important output features despite these underlying instabilities. These processes are poorly understood, and will be examined here by exploitation of a well characterized pattern-generating neural network, the crustacean stomatogastric ganglion (STG). The STG contains the motor neurons of two networks that drive rhythmic motor patterns involved in the processing of food: the pyloric (filtering) network, which is constitutively active both in vivo and in disseced in vitro preparations of the stomatogastric nervous system; and the gastric (chewing) network, which is episodic, being activated by modulatory inputs only after feeding. Previous in vitro work has shown that upon removal of these inputs by transection of the only input nerve (decentralization), both types of rhythmic activity cease, but after a period of several days in culture, rhythmic pyloric activity returns. Gastric activity has not been previously described in decentralized preparations, but preliminary data indicates that gastric recovery may also occur in this paradigm. This work will characterize changes occurring in the gastric network following days of decentralization, when maintained in organotypic culture, and test the hypothesis that the gastric network can also regain its rhythmic character, despite the loss of modulatory input. Unlike the pyloric network, the gastric circuit has been shown to directly involve the nerve terminals of descending modulatory inputs; thus any recovery must occur by a reorganization of the essential circuit, a fundamentally different process than that which has been described in the pyloric network. This hypothesis will be tested by three main avenues. First, the phenomenology of recovery will be characterized by continuously recording gastric activity extracellularly along output motor nerves. Second, intracellular changes occurring during the course of recovery in gastric cells will be determined by measuring a variety of intrinsic properties. Third, the abilityof severed neuromodulatory inputs to participate in recovered circuitry will be tested by stimulation and immunocytochemical staining. Taken together, these experiments will provide a window into the complex regulatory processes involved in the maintenance of stable network function. Specifically, this system potentially offers an excellent model for the changes that occur in vertebrate spinal cord injury, where the motor patterns involved in locomotion are also episodic, and frequently develop pathologic activity following loss of modulatory input.